Conventionally, a method is disclosed which is configured to: compare an engine rotational speed with a clutch input rotational speed when a vehicle start clutch is engaged gradually; and learn as a torque transmission point a clutch supply hydraulic pressure when the clutch input rotational speed falls by a predetermined rotational speed with respect to the engine rotational speed (see a patent document 1, for example).
A lockup clutch provided in a torque converter is also required to strike a balance between quick engagement for enhancement of fuel efficiency and suppression of driver's uncomfortable feeling due to change of vehicle behavior at engagement. For a solution, it is conceivable that a torque transmission point of the lockup clutch is also learned, and the learning control for the vehicle start clutch is applied to the lockup clutch.
However, if for ensuring the frequency of learning, the learning control for the torque transmission point of the lockup clutch is performed while the vehicle is running, it is possible that the engine rotational speed is changed by a factor other than the state of engagement of the lockup clutch, causing erroneous learning.
Furthermore, a general learning control, in which when a calculated detection error of a learning detection value is above a predetermined value, a constant learning value correction quantity is added to a last learning value to obtain a present learning value, is confronted by the following problem. If the learning value correction quantity is set to a large constant value, in a situation where the last learning value is close to a true value after convergence and the present detection error is based on an erroneous learning detection value, the present learning value becomes apart from the true value. On the other hand, if the learning value correction quantity is set to a small constant value, in a situation where the last learning value is apart from the true value, multiple learning experiences are required for convergence of the learning value to the true value.
The present invention is made with attention to the problem described above, and is targeted for providing a vehicular lockup clutch control device and a vehicular lockup clutch control method capable of allowing a learning value to converge to and be held at a true value, while ensuring the frequency of learning, wherein a learning control is performed based on information about a meeting point at which a lockup clutch starts torque transmission.